skill-distiller

To decide if workflow outcomes should become reusable skills, evaluate the material for recurrence, non-obviousness, and generality. This process gates candidate workflows, preventing skill proliferation while capturing genuinely valuable automation logic.

Organizing knowledge management outputs from planning sessions involves flagging potential material, assessing its significance, and checking for redundancy. You then either refine it into a new SKILL.md file or fold it into an existing skill to maintain a clean library.

Yes, you can fold new automation logic into an existing skill to avoid duplication. The workflow checks candidate material against your current skill library, suggesting consolidation when overlap is high rather than authoring a redundant SKILL.md file.

Assessing non-obviousness and generality when creating skills from conversations involves a five-step workflow: flag the candidate material, evaluate its significance and recurrence, check redundancy against existing skills, refine the logic, and emit the final decision.

You should avoid creating a new skill from plan reflections when the material lacks recurrence or generality, or when existing skills already cover the workflow. Queuing promising but premature candidates for later authoring prevents unnecessary skill proliferation.